In recent years, we entered the age of multimedia in which voices, images and other representative media are integratedly operated, and means for transmitting information of conventional information media, such as newspapers, magazines, television, radio and telephone, to persons have been adopted as targets of the multimedia. The multimedia generally means representing not only characters but also drawings, voices, particularly images and the like, which are simultaneously correlated with the characters. However, in order to make the conventional information media the targets of the multimedia, it is indispensable to represent the information in digital formats.
Estimating the amount of the information in each of the information media as a digital information amount, the information amount of characters (per character) is 1˜2 bytes, the information amount of voices of telephone quality is 64 Kbits per seconds and the information amount of moving pictures of the current television receiving quality is 100 Mbits per second. Accordingly, in information media such as the telephone and television, it is not practical that massive information in digital formats is operated as it is. For example, TV phones have already been put to practical use by ISDN (Integrated Service Digital Network) having the transmission rate of 64 kbps˜1.5 Mbps. However, video information which is obtained by TV cameras cannot be transmitted as digital data as it is by the ISDN.
Thus, the technology for compressing information is required. For example, in the case of TV phones, the moving picture compression technology according to H.261 and H.263 standards which are internationally standardized by ITU-T (International Telecommunication Union-Telecommunication Sector) is employed. In addition, according to the information compression technology of MPEG-1 standard, image information can be stored in normal music CDs (Compact Disks) together with voice information.
Here, MPEG (Moving Picture Experts Group) is the international standard relating to the processing for compressing moving picture data. MPEG-1 is the standard for compressing moving picture data in up to 1.5 Mbps, i.e., compressing information of television signals by about one hundredth. Further, the transmission rate for MPEG-1 standard is principally restricted to about 1.5 Mbps. The moving picture data are compressed into 2˜15 Mbps according to MPEG-2 which is standardized to meet requirements of further higher image quality.
Further, in the present circumstances, MPEG-4 which enables coding of image data and operation of image data in object units, and realizes new functions required in the age of multimedia is now being standardized by a working group (ISO/IEC JTC1/SC29/WG11) that has promoted the standardization of the compression processing for moving picture data, like MPEG-1 and MPEG-2. In MPEG-4, the standardization of the coding processing at a lower bit rate has been aimed at first, but at the present time the targets of the standardization are extended to general-purpose coding processing for interlaced images at a higher bit rate.
One of characteristics of MPEG-4 is a mechanism for simultaneously coding image data corresponding to plural image sequences (i.e., plural moving pictures) and transmitting them. This mechanism makes it possible to construct one scene by composing plural images. The image in this case is an image (still picture) of each frame of the image sequence (moving picture). One scene is a composed image including plural images.
For example, in MPEG-4, it is possible that the foreground and background constituting one scene are separated as images (objects) of different image sequences and that the frame frequency, image quality and bit rate are changed independently for each of the image sequences. In addition, in MPEG-4, the images of the plural image sequences are arranged in the horizontal or vertical direction, like multi-screens, whereby users can extract or enlarge to display only images of desired image sequences.
As for the background, the coding processing only for pixel value signals (texture signals) indicating the brightness and tint is generally performed, as in MPEG-2. On the other hand, as for the foreground, the processing of not only coding the pixel value signals indicating the brightness and tint of the objects but also simultaneously coding shape signals indicating shapes of the objects is performed. Generally, this coding processing for the foreground is known as coding processing in object units.
According to MPEG-4, the whole of a displayed image (composed image) is composed of images (objects) of plural image sequences. Therefore, a frame of each image sequence at each display time is referred to as a VOP (Video Object Plane) and distinguished from frames according to MPEG-1 and 2. When the whole of the displayed image is composed of images of one image sequence, the VOP and the frame coincide.
FIG. 8 is a diagram FIGS. 8(a) to 8(c) are diagrams for schematically explaining coding processing in object units according to MPEG-4.
An image signal specified according to MPEG-4 is composed of a shape signal representing the shape Sob of an object (VOP) (FIG. 8(a)) and a pixel value signal (texture signal) comprising a brightness signal and a color difference signal, representing the texture Tob of the object (VOP) (FIG. 8(b)).
In the coding processing in object units, it is necessary to decide the shape of the object and the position of the object with respect to a reference coordinate system for displaying the image. Thus, a rectangular region (bounding box) Box (FIG. 8(c)) is composed of plural macroblocks, involving the object Ob, is decided by the reference coordinate system. Here, the macroblock is an image space as a unit of the coding processing, and composed of 16×16 pixels. In addition, since the rectangular region Box is composed of the plural macroblocks, the number of pixels in the horizontal and vertical directions in the rectangular region is a multiple of 16.
Then, each of the rectangular regions Box in one image sequence is subjected to the coding processing of coding the image signal for each macroblock.
For example, in FIG. 8(c), the rectangular region Box is composed of 5×4 macroblocks. Macroblocks MB1 and MB2 are macroblocks situated outside the object Ob (macroblock-outside-object). A macroblock MB3 is a boundary macroblock situated on the boundaries of the object Ob. A macroblock MB13 is a macroblock situated inside the object Ob (macroblock-inside-object). According to MPEG-4, pixels outside the object are not displayed after the decoding. Therefore, the coding processing is performed only for macroblocks including the pixels inside the object, which are displayed after the decoding, i.e., the boundary macroblocks and macroblocks-inside-object.
FIG. 9 is a diagram FIGS. 9(a) to 9(e) are diagrams for schematically explaining various processing units in a bitstream which conforms to MPEG-4.
Here, a rectangular region (bounding box) Box including an object (VOP) corresponds to the object in a one-to-one relationship. Therefore, in the following description, the rectangular region (bounding box) Box and the object (VOP) are not distinguished from each other and they are referred to as VOPs.
Generally, in a code sequence (bitstream) composed of variable length codes, fixed length codes comprising specific bit patterns are arranged to prevent the error propagation at the decoding time. According to MPEG-4, this fixed length code is referred to as Resync Marker (hereinafter abbreviated as only marker) and it is a synchronization signal. A code sequence composed of this marker and the variable length codes subsequent to the marker is a coding unit which is referred to as a video packet.
According to MPEG-4, as shown in FIG. 9(a), a code sequence (VOP bitstream) Svop corresponding to one VOP 10 can be composed of plural video packets. In this case, the VOP bitstream Svop is composed of four video packets Svp1˜Svp4. Coded data corresponding to respective regions Rvp1˜Rvp4 in the VOP 10 are stored in the video packets Svp1˜Svp4, respectively. In addition, coded data corresponding to plural macroblocks can be stored in each of the video packets Svp1˜Svp4.
Here, the region Rvp1 corresponding to the video packet Svp1 is composed of five macroblocks MB1˜MB5 as shown in FIG. 9(b). Also each of the regions Rvp2˜Rvp4 corresponding to the other video packets Svp2˜Svp4 is composed of five macroblocks as the region Rvp1 corresponding to the video packet Svp1. In addition, each of the macroblocks is an image space composed of 16 pixels×16 pixels as described above and composed of four blocks. Each of the blocks is an image space composed of 8 pixels×8 pixels. For example, the macroblock MB1 is composed of blocks B1˜B4 as shown in FIG. 9(c), and the block B1 is composed of 8 pixels×8 pixels as shown in FIG. 9(d).
In addition, the coded data corresponding to one macroblock (hereinafter referred to also as macroblock information) include brightness information (Y) which corresponds to the four blocks constituting one macroblock, and color difference information (U) and (V) corresponding to one macroblock. Further, when the object has the shape, the coded data of one macroblock include shape information corresponding to one macroblock, together with the brightness information (Y) and the color difference information (U) and (V).
Here, the brightness information (Y) of one macroblock is obtained by coding the pixel value signals of the four blocks constituting one macroblock. The color difference information (U) and (V) of one macroblock is obtained by coding the color difference signals (U) and (V) of 8 pixels×8 pixels constituting one macroblock, respectively. The shape information of one macroblock is obtained by coding the shape signal of 16 pixels×16 pixels constituting one macroblock.
It is unnecessary that the number of the macroblocks constituting the region corresponding to the video packet in the VOP 10 is always fixed as shown in FIG. 9(a). For example, as shown in FIG. 9(e), the numbers of macroblocks constituting regions Rvp1a˜Rvp5a which correspond to video packets in a VOP 10a may be decided so that the amounts of codes in the respective video packets Svp1a˜Svp5a in a VOP bitstream Svopa are fixed. In this case, the numbers of the macroblocks included in the regions Rvp1a˜Rvp5a corresponding to the respective video packets are not fixed.
FIG. 10 is a diagram for schematically explaining the coding processing according to MPEG-4 in object units. This figure shows the coding processing for the image signal corresponding to the object (VOP) having the shape as shown in FIG. 8.
Here, the VOP 10 as the object Ob (strictly speaking Bounding Box BBOX including the object) is composed of four video packet regions Rvp1˜Rvp4 each composed of five macroblocks. For example, the video packet region Rvp1 is composed of the macroblocks MB1˜MB5.
The macroblocks MB1 and MB2 are situated outside the object. Therefore, these macroblocks MB1 and MB2 are subjected to the processing of coding the shape signal indicating that the macroblocks are outside the object as the coding processing for the shape signal, and the processing of coding the pixel value signal is omitted. In addition, the macroblock MB3 is subjected to the coding processing for the shape signal and the coding processing for the pixel value signal, because this macroblock is a macroblock including pixels inside the object.
Generally, many objects having the shapes as the foreground have the shapes or sizes varying from moment to moment, unlike objects as the background. In addition, according to MPEG-4, the coding algorithm of the shape signals or pixel value signals greatly depends on the shapes of the images to be coded. For example, when an object has the shape, the coding processing of the pixel value signal is omitted for a part (macroblock) which is indicated by the shape signal that it is outside the object. Therefore, there are some cases where the number of the macroblocks which are subjected to the coding processing of the pixel value signals, corresponding to one scene (VOP) of one image sequence may vary. Accordingly, the decoding processing conforming to MPEG-4 is easily affected by transmission errors of the bitstream, as compared with the decoding processing for the coding processing of images having the shapes and sizes which do not vary like MPEG-2. Further, in this decoding processing, the image concealment such as the image restoration or image processing utilizing the correlation between VOPs is also difficult. Consequently, in the decoding system for MPEG-4, when the transmission error occurs, the image quality in a decoded image is considerably deteriorated.
FIGS. 11(a) to 11(c) are diagrams for schematically explaining the structure of a bitstream which conforms to MPEG-4 in detail.
The VOP bitstream Svop includes coded data corresponding to the VOP 10 as the object Ob as shown in FIG. 10. At the head of the VOP bitstream Svop, a VOP header Svoph as important data relating to the whole VOP is arranged. The video packets Svp1˜Svp4 are arranged subsequent to the VOP header Svoph (see FIG. 11(a)).
In the video packet Svp1, a video packet header Svph as important data relating to the whole video packet is arranged at its head. Coded data (macroblock information) Smb1˜Smb5 corresponding to the macroblocks MB1˜MB5 are arranged subsequent to the video packet header Svph (see FIG. 11 (b)).
Further, a macroblock header Smbh as important data relating to the whole macroblock is arranged at the head of the macroblock information Smb1. Shape information Ssb, of the corresponding macroblock, brightness information Spb1˜Spb4 of four blocks constituting the corresponding macroblock, and color difference information (U) Spbu and (V) Spbv of the corresponding macroblock are arranged subsequent to the macroblock header Smbh (see FIG. 11(c)).
Thus, in the VOP bitstream Svop, the macroblock information corresponding to the macroblock as the coding unit is the first processing unit. Further, the video packet composed of the plural pieces of macroblock information is the second processing unit. The VOP bitstream has a two-layer data structure in which the coded data included therein are divided by the first and second processing units.
Here, the VOP header Svoph and the video packet header Svph include synchronization signals for synchronizing the decoding processing of the bitstream. Thus, when the decoding processing of the bitstreams is interrupted due to an error bit in the bitstream, the decoding processing can be resumed from the VOP header Svoph or video packet header Svph. On the other hand, the macroblock header Smbh includes no synchronization signal for synchronizing the decoding processing. Here, the synchronization signal in the video packet header Svph is the above-mentioned fixed length code (Resync Marker).
Generally, as errors in the bitstream in the moving picture decoding processing, there are two kinds of errors, i.e., a stream error and a transmission error.
As the stream error, there is an error that a code which is grammatically incorrect is included in a stream (syntax error) or an error that a code of an incorrect value which exceeds a range of available values is included (semantic error) and the like. The transmission error is an error in which the bitstream is destroyed when the bitstream is read from a recording medium or the bitstream is transmitted via a communication medium due to the missing of data or the like.
Usually, the coded image data corresponding to each VOP are stored in a transmission packet having header information, and transmitted as the VOP bitstream in transmission packet units. Therefore, when the transmission error such as the absence of packets occurs, the position of a missing transmission packet in the bitstream can be detected on the receiving end. Accordingly, as for the transmission error, the position where the decoding processing fails in the bitstream (error occurrence position) can be almost specified.
As a concrete method for specifying the error occurrence position in the decoding processing, there is a method of detecting the absence of packets in the bitstream and adding a mark which indicates the absence of the packet (marker code) to the position of the missing packet in the bitstream.
Compared with this transmission error, the stream error results from the syntax error occurring at the variable-length coding time or the like. Therefore, this error cannot be detected as a decoding error until the decoding process such as the variable-length decoding processing fails. In other words, the stream errors cannot be detected essentially unless the decoding process of the bitstream (coded data) fails.
However, the synchronization signal is arranged at the head of one video packet. In addition, immediately after this video packet, the synchronization signal of the subsequent video packet is arranged. Therefore, when the structure and contents of the bitstream situated between these two synchronization signals are strictly examined in the decoding process, only video packets including the stream errors can be detected regardless of the failure of the decoding process of the bitstream. When the structure and contents of the bitstream are strictly examined in the decoding process in this way, the possibility of detecting the stream errors is considerably higher as compared with the case where the failure of the decoding process of the bitstream is detected.
Hereinafter, a conventional moving picture decoding apparatus will be described in detail.
FIG. 12 is a block diagram for explaining the conventional common moving picture decoding apparatus.
This moving picture decoding apparatus 100 receives a bitstream read from a recording medium or a bitstream transmitted via a transmission medium as an input stream Bin and performs decoding processing for the input stream Vin. Here, the bitstream includes image coded data which are obtained by subjecting an image signal of a moving picture to coding processing separately for each of plural image sequences constituting the moving picture. In addition, the coding processing for the image signal of one of the image sequences is performed for each scene (VOP) of the image sequence and the image signal corresponding to each VOP is coded in units of macroblocks constituting the VOP. It goes without saying that the image signal of the object having no shape includes only the brightness signal and the color difference signal, and the image signal of the object having the shape includes the shape signal together with the brightness signal and the color difference signal.
The bitstream corresponding to the moving picture normally includes the image coded data corresponding to each object being multiplexed. However, in the following description, assume that the bitstream includes only image coded data corresponding to one object as image information.
To be specific, the moving picture decoding apparatus 100 includes a decoder 101 for performing the decoding processing of the input stream Vin corresponding to a target VOP to be processed for each macroblock with reference to decoded image data (reference image data) Vref in a reference region in an already processed VOP whose decoding processing is finished, and outputting decoded image data Vd, and a memory 102 for synchronizing the reference image data Vref with decoding processing of a macroblock to be processed (target macroblock) in the target VOP, and outputting the data as well as synchronizing decoded image data (replacement image data) Vrep corresponding to a macroblock in the already processed VOP whose relative position in the already processed VOP is equal to the relative position of the target macroblock in the target VOP with the decoding processing for the target macroblock and outputting the data.
Further, the moving picture decoding apparatus 100 includes an error detector 120 for detecting an error of the input stream Vin and its position on the basis of the input stream Vin and outputting an error notification signal Terr, a selector switch 105 for selecting one of the decoded image data Vd of the target macroblock and the replacement image data Vrep in accordance with a control signal Cmb, and outputting the selected image data (MB selected image data) Emb as reproduced image data Vout of the target macroblock, and a macroblock unit concealer 104 for generating the control signal Cmb for the selector switch 105 in accordance with the error notification signal Terr.
Here, the error detector 120 detects the error in the input stream Vin using the level of the input stream Vin as an analog signal or an error-correcting signal included in the input stream. Therefore, in this error detector 120, the transmission errors are detected.
The macroblock unit concealer 104 controls the selector switch 105 in accordance with the error notification signal Terr so that in place of the decoded image data Vd which are obtained by decoding the macroblock information from the macroblock information including the error part of the input stream Vin to the subsequent synchronization signal, the decoded image data (replacement image data) Vrep of the already processed VOP corresponding to the decoded image data are output as the reproduced image data Vout.
Next, the operation of the moving picture decoding apparatus 100 is described.
When a bitstream read from the recording medium or a bitstream transmitted via the transmission medium is input to the moving picture decoding apparatus 100 as the input stream Vin, the decoding processing for the input stream is performed in macroblock units for each VOP in the moving picture decoding apparatus 100. Here, in this moving picture decoding apparatus 100, the decoder 101, the memory 102 and the macroblock unit concealer 104 are controlled by a control unit (not shown) in this apparatus 100 during the decoding processing so that the processings for the respective macroblocks are synchronized between these units.
To be specific, in the decoder 101, the decoding processing is performed for the coded data of the target macroblock in the target VOP, with reference to the reference image data Vref corresponding to the target macroblock, and the decoded image data Vd of the target macroblock are output. When the input stream Vin includes an error, the decoder 101 outputs only the decoded image data Vd corresponding to the macroblock whose coded data can be decoded.
At this time, the memory 102 outputs the replacement image data Vrep corresponding to the target macroblock together with the reference image data Vref corresponding to the target macroblock.
The error detector 120 performs error detection processing of detecting transmission errors on the basis of the input stream Vin. When the error of the input stream is detected, the error detector 120 outputs an error notification signal Terr which shows the macroblock information including an error part of the input stream as the position of the error part, to the macroblock unit concealer 104.
Then, the macroblock unit concealer 104 outputs the control signal Cmb to the selector switch 105 for selecting one of the decoded image data Vd of the target macroblock and the replacement image data Vrep corresponding to the target macroblock, in accordance with the error notification signal Terr. That is, the selector switch 105 is controlled so that the replacement image data Vrep from the memory 102 are selected, in place of the decoded image data Vd from the decoder 101, for the macroblock corresponding to each of macroblock information from the macroblock information indicated by the error notification signal Terr to the subsequent synchronization signal, and the decoded image data Vd output by the decoder 101 are selected for other macroblocks.
Then, the selected image data Emb which are selected by the selector switch 105 are output as the reproduced image data Vout corresponding to the target macroblock of the target VOP. In addition, the selected image data Emb are recorded in the memory 102 as the reference image data for a VOP subsequent to the target VOP.
At this time, not only the decoded image data of the error macroblock (macroblock whose macroblock information includes the error part of the bitstream) but also the decoded image data of all macroblocks subsequent to the error macroblock in the video packet are replaced with the decoded image data (replacement image data) Vrep of the corresponding macroblocks in the already processed VOP, because the input stream is obtained by the variable-length coding processing for the image data. To be specific, in the variable-length decoding processing for the input stream, when the input stream includes an error, the error affects the decoding processing for all of macroblock information from the error occurrence position in the input stream to the synchronization signal.
FIG. 13 is a block diagram for explaining another conventional moving picture decoding apparatus.
This moving picture decoding apparatus 110 conceals the decoded image data which are obtained by the decoding processing for the input stream which includes errors, not in macroblock units as in the moving picture decoding apparatus 100 but in video packet units.
To be specific, this moving picture decoding apparatus 110, like the moving picture decoding apparatus 100 shown in FIG. 12, has a decoder 101 for performing the decoding processing for the input stream Vin corresponding to the target VOP with reference to the reference image data Vref and outputting the decoded image data Vd corresponding to each macroblock, and a memory 102 for synchronizing the reference image data Vref and the replacement image data Vrep for the target macroblock with the decoding processing of the target macroblock and outputting the data.
Further, in place of the selector switch 105 in the moving picture decoding apparatus 100, the moving picture decoding apparatus 110 has a first delay circuit 103 for delaying the decoded image data Vd for a time required for the decoding processing of a target video packet to be processed, a second delay circuit 104 for delaying the replacement image data Vrep which are output by the memory 102 in synchronization with the decoding processing of each macroblock, for a time required for the decoding processing of the target video packet, and a selector switch 108 for selecting one of the output (delayed decoded data) DVd of the first delay circuit 103 and the output (delayed replacement data) DVrep of the second delay circuit 104, in accordance with a control signal Cvp.
Further, in place of the error detector 120 in the moving picture decoding apparatus 100, this moving picture decoding apparatus 110 has an error detector 121 for detecting the failure of the normal decoding processing in the decoder 101 in accordance with an internal signal Si of the decoder, and outputting an error notification signal Nerr indicating the error detection. This error detector 121 can have a structure for detecting the abnormality of the bitstream in the video packet by the processing of strictly examining the structure and contents of the bitstream, in place of the processing of detecting the failure of the normal decoding processing, and outputting the error notification signal Nerr indicating the error detection.
Further, in place of the macroblock unit concealer 104 in the moving picture decoding apparatus 100, this moving picture decoding apparatus 110 has a video packet unit concealer 107 for controlling the selector switch 108 to select one of the delayed decoded data DVd from the first delay circuit 103 and the delayed replacement data DVrep from the second delay circuit 104 for each macroblock, in accordance with the error notification signal Nerr.
To be specific, this video packet unit concealer 107 controls the selector switch 108 in accordance with the error notification signal Nerr so as to output, in place of the delayed decoded data DVd corresponding to a video packet (error video packet) whose decoding processing in the decoder 101 fails, the delayed replacement data DVrep of a video packet of the already processed VOP corresponding to the error video packet as the reproduced image data Vout.
Here, the decoder 101 and the memory 102 in the moving picture decoding apparatus 110 as shown in FIG. 13 have the same structures as those of the decoder 101 and the memory 102 in the moving picture decoding apparatus 100 as shown in FIG. 12.
Next, the operation of the moving picture decoding apparatus 110 is described.
In this moving picture decoding apparatus 110, the decoding processing for the input stream Vin in the decoder 101, and outputting of the reference image data Vref and the replacement image data Vrep from the memory 102 is performed in the same way as in the moving picture decoding apparatus 100.
In this moving picture decoding apparatus 110, the decoded image data Vd from the decoder 101 are delayed by the first delay circuit 103 for a time required for the decoding processing for a target video packet to be decoded, and the replacement image data Vrep from the memory 102 are delayed by the second delay circuit 104 for a time required for the decoding processing for the target video packet.
In the error detector 121, the processing of detecting the failure of the decoding processing for the input stream is performed in accordance with the internal signal Si in the decoder 101. When the failure of the decoding processing is detected, the error notification signal Nerr indicating the detection of the error is output to the video packet unit concealer 107. This video packet unit concealer 107 outputs the control signal Cvp to the selector switch 108 in accordance with the error notification signal Nerr. The selector switch 108 selects one of the delayed decoded data DVd from the first delay circuit 103 and the delayed replacement data DVrep from the second delay circuit 104 in accordance with the control signal Cvp, and outputs the selected data (VP unit selected data) Evp as the reproduced image data Vout.
To be more specific, the selector switch 108 is controlled by the video packet unit concealer 107 so that, in place of the delayed decoded data DVd corresponding to the video packet in which the error is detected (error video packet), the delayed replacement data DVrep of the video packet of the already processed VOP, corresponding to the error video packet, are output as the reproduced image data Vout.
The reproduced image data Vout of the target VOP are recorded in the memory 102 as the reference image data for the VOP subsequent to the target VOP.
The so-constructed moving picture decoding apparatus 110 detects the failure of the decoding processing, and replaces the decoded image data Vd of the video packet whose decoding processing fails, with the decoded image data of the corresponding video packet in the already processed VOP. The decoding processing normally fails when the input stream includes errors. Therefore, when the bitstream including the transmission errors or stream errors is input, the concealment of the decoded image data is performed.
However, the above-mentioned conventional moving picture decoding apparatus, i.e., the conventional moving picture decoding apparatus 100 which conceals the decoded image data in macroblock units (see FIG. 12) and the conventional moving picture decoding apparatus 110 which conceals the decoded image data in video packet units (see FIG. 13) have following problems, respectively.
To be specific, the moving picture decoding apparatus 100 as shown in FIG. 12 detects errors using the analog signal level of the input stream or the error-correcting codes, and conceals the decoded image data in macroblock units, whereby the decoded image data can be concealed carefully. However, the stream errors cannot be detected using the analog signal level of the input stream or the error-correcting codes. Therefore, the deterioration in image quality of the decoded image due to the stream errors cannot be reduced.
The moving picture decoding apparatus 110 as shown in FIG. 13 detects errors on the basis of the occurrence of the failure in the decoding processing, and conceals the decoded image data in video packet units. Therefore, the decoded image data corresponding to the normal macroblock information including no error part, precedent to the error macroblock are also replaced with the decoded image data of the already processed VOP. Thus, the image quality of the decoded image is considerably deteriorated due to the concealment of the decoded image data and thereby the concealment of the transmission errors or stream errors in the decoded image cannot be performed effectively.
Therefore, conventionally, the moving picture decoding apparatus which conceals the decoded image in macroblock units as shown in FIG. 12 and the moving picture decoding apparatus which conceals the decoded image in video packet units as shown in FIG. 13 are used properly according to their purposes.
Further, both of the above-mentioned conventional moving picture decoding apparatuses have the structures in which the concealment processing for the decoded image is performed without distinction between the case where the input stream has the shape information and the case where the input stream has no shape information. Therefore, in the case where the input stream has the shape information, a good image quality is not obtained even when the concealment of the image is performed.
That is, there are many cases where objects as the targets of the coding processing in object units have the shapes which considerably vary from moment to moment. Therefore, when the image of a part of the target VOP is concealed utilizing an image in the already processed VOP, the continuity in shape within the target VOP is often harmed between concealed parts and unconcealed parts in the target VOP. When the continuity in shape is harmed, the concealed parts show and the image quality is substantially deteriorated.